Related Experiment Video
Updated: Jun 14, 2025
10:08
Co-culture of Glioblastoma Stem-like Cells on Patterned Neurons to Study Migration and Cellular Interactions
Published on: February 24, 2021
5.9K
Graph-based cell pattern recognition for merging the multi-modal optical microscopic image of neurons.
Wenwei Li1, Wu Chen1, Zimin Dai1
1Britton Chance Center for Biomedical Photonics, Wuhan National Laboratory for Optoelectronics, MoE Key Laboratory for Biomedical Photonics, Huazhong University of Science and Technology, Wuhan 430074, PR China.
Computer Methods and Programs in Biomedicine
|September 3, 2024
Summary
This study presents a novel graph-model approach for accurately matching neurons across different imaging types, crucial for understanding brain circuitry and function.
Area of Science:
- Neuroscience
- Biophysics
- Computational Biology
Background:
- Understanding neuron structure and function is vital for brain research and disease diagnosis.
- Optical microscopy is essential for visualizing neuronal morphology and activity.
- Simultaneously capturing dynamic in vivo and static ex vivo neuronal data is challenging due to positional displacement.
Purpose of the Study:
- To develop an automated method for precise cell image matching of sparsely labeled neurons across multimodal optical microscopic images.
- To enable the integration of functional and structural imaging data at the single-neuron level.
- To overcome challenges in aligning neuron positions between in vivo and ex vivo imaging.
Main Methods:
- A graph-model-based approach for automated cell image matching.
- Utilizing neuron distribution as a matching feature to mitigate modal differences.
- Employing a high-order graph model to address scale inconsistency.
- Applying nonlinear iteration to resolve discrepancies in neuron density.
Main Results:
- Achieved 96.67% precision, 85.29% recall rate, and 90.63% F1 Score in cell matching.
- Demonstrated performance comparable to expert technicians.
- Successfully applied the strategy to a connectivity study in the mouse visual cortex, matching two-photon calcium images with HD-fMOST brain-wide anatomical images.
Conclusions:
- The developed graph-model approach provides precise and automated neuron pairing across different imaging modalities.
- This method bridges the gap between functional and structural imaging, crucial for neuron classification and circuitry analysis.
- Offers significant technical support for advanced neuroscience research.
